A bundle Bregman proximal method for convex nondifferentiable minimization

k } by taking x^k to be an approximate minimizer of , where is a piecewise linear model of f constructed from accumulated subgradient linearizations of f, D_h is the D-function of a generalized Bregman function h and t_k>0. Convergence under implementable criteria is established by extending our recent framework of Bregman proximal minimization, which is of independent interest, e.g., for nonquadratic multiplier methods for constrained minimization. In particular, we provide new insights into the convergence properties of bundle methods based on h=?|·|². Received September 18, 1997 / Revised version received June 30, 1998 Published online November 24, 1998     

Linear convergence of epsilon-subgradient descent methods for a class of convex functions

This paper establishes a linear convergence rate for a class of epsilon-subgradient descent methods for minimizing certain convex functions on ℝ ⁿ . Currently prominent methods belonging to this class include the resolvent (proximal point) method and the bundle method in proximal form (considered as a sequence of serious steps). Other methods, such as a variant of the proximal point method given by Correa and Lemaréchal, can also fit within this framework, depending on how they are implemented. The convex functions covered by the analysis are those whose conjugates have subdifferentials that are locally upper Lipschitzian at the origin, a property generalizing classical regularity conditions. Received March 29, 1996 / Revised version received March 5, 1999? Published online June 11, 1999     

Convergence Analysis of Some Methods for Minimizing a Nonsmooth Convex Function

In this paper, we analyze a class of methods for minimizing a proper lower semicontinuous extended-valued convex function . Instead of the original objective function f, we employ a convex approximation f _{k + 1} at the kth iteration. Some global convergence rate estimates are obtained. We illustrate our approach by proposing (i) a new family of proximal point algorithms which possesses the global convergence rate estimate even it the iteration points are calculated approximately, where are the proximal parameters, and (ii) a variant proximal bundle method. Applications to stochastic programs are discussed.     

Sufficient descent nonlinear conjugate gradient methods with conjugacy condition

In this paper, by the use of Gram-Schmidt orthogonalization, we propose a class of modified conjugate gradient methods. The methods are modifications of the well-known conjugate gradient methods including the PRP, the HS, the FR and the DY methods. A common property of the modified methods is that the direction generated by any member of the class satisfies g_k^Td_k=-||g_k||²g_{k}^{T}d_k=-\|g_k\|^2. Moreover, if line search is exact, the modified method reduces to the standard conjugate gradient method accordingly. In particular, we study the modified YT and YT+ methods. Under suitable conditions, we prove the global convergence of these two methods. Extensive numerical experiments show that the proposed methods are efficient for the test problems from the CUTE library.     

An interior point multiplicative method for optimization under positivity constraints

We analyze an algorithm for the problem minf(x) s.t.x 0 suggested, without convergence proof, by Eggermont. The iterative step is given by x _j ^k+1 =x _j ^k (1-_kf(x ^k)_j) with _k > 0 determined through a line search. This method can be seen as a natural extension of the steepest descent method for unconstrained optimization, and we establish convergence properties similar to those known for steepest descent, namely weak convergence to a KKT point for a generalf, weak convergence to a solution for convexf and full convergence to the solution for strictly convexf. Applying this method to a maximum likelihood estimation problem, we obtain an additively overrelaxed version of the EM Algorithm. We extend the full convergence results known for EM to this overrelaxed version by establishing local Fejér monotonicity to the solution set.Research for this paper was partially supported by CNPq grant No 301280/86.     

Some Convergence Properties of Descent Methods

In this paper, we discuss the convergence properties of a class of descent algorithms for minimizing a continuously differentiable function f on R ⁿ without assuming that the sequence { x _k } of iterates is bounded. Under mild conditions, we prove that the limit infimum of is zero and that false convergence does not occur when f is convex. Furthermore, we discuss the convergence rate of { } and { f(x _k )} when { x _k } is unbounded and { f(x _k )} is bounded.     

An infeasible-interior-point potential-reduction algorithm for linear programming

n . The method is based on Rockafellar’s proximal point algorithm and a cutting-plane technique. At each step, we use an approximate proximal point p^a(x_k) of x_k to define a v_k∈∂_εkf(p^a(x_k)) with ε_k≤α∥v_k∥, where α is a constant. The method monitors the reduction in the value of ∥v_k∥ to identify when a line search on f should be used. The quasi-Newton step is used to reduce the value of ∥v_k∥. Without the differentiability of f, the method converges globally and the rate of convergence is Q-linear. Superlinear convergence is also discussed to extend the characterization result of Dennis and Moré. Numerical results show the good performance of the method. Received October 3, 1995 / Revised version received August 20, 1998 Published online January 20, 1999     

On the 2<Emphasis Type="Italic">m</Emphasis>-th power mean of Dirichlet <Emphasis Type="Italic">L</Emphasis>-functions with the weight of trigonometric sums

Let p be a prime, χ denote the Dirichlet character modulo p, f (x) = a ₀ + a ₁ x + ... + a _k x ^k is a k-degree polynomial with integral coefficients such that (p, a ₀, a ₁, ..., a _k ) = 1, for any integer m, we study the asymptotic property of

Generalized uncertainty inequalities

The Heisenberg–Pauli–Weyl (HPW) uncertainty inequality on \mathbbRⁿ{\mathbb{R}^n} says that

White noise eliminates instability

Let x₁,..., x_n be points in the d-dimensional Euclidean space E^d with || x_i-x_j|| ￡ 1\| x_{i}-x_{j}\| \le 1 for all 1 \leqq i,j \leqq n1 \leqq i,j \leqq n, where || .||\| .\| denotes the Euclidean norm. We ask for the maximum M(d,n) of \mathop?_i, j=1ⁿ|| x_i-x_j|| ²\textstyle\mathop\sum\limits _{i,\,j=1}^{n}\| x_{i}-x_{j}\| ^{2} (see [4]). This paper deals with the case d = 2. We calculate M(2, n) and show that the value M(2, n) is attained if and only if the points are distributed as evenly as possible among the vertices of a regular triangle of edge-length 1. Moreover we give an upper bound for the value \mathop?_i, j=1ⁿ|| x_i-x_j|| \textstyle\mathop\sum\limits _{i,\,j=1}^{n}\| x_{i}-x_{j}\| , where the points x₁,...,x_n are chosen under the same constraints as above.     

A note on symbolic and ordinary powers of homogeneous ideals

In this note we are interested in the graded modulesM _k=I^(k)/I^k and , whereI is a saturated ideal in the homogeneous coordinate ringS=K[x₀,…,x_n] of ℙⁿ,I ^(k) is the symbolic power and is the saturation of the ordinary power. Very little is known about these modules, and we provide a bound on their diameters, we compute the Hilbert functions and we study some characteristic submodules in the special case ofn+1 general points in ℙⁿ.

---

Sunto In questa nota siamo interessati ai moduli graduatiM _k=I(k)/I^k e , doveI è un ideale saturato nell'anello delle coordinate omogeneeS:=K[x₀,…,x_n] di ℙⁿ,I ^(k) è la potenza simbolica e è la saturazione della potenza ordinaria. Poco è noto su questi moduli e qui viene fornito un limite superiore ai loro diametri. Ne calcoliamo inoltre le funzioni di Hilbert e studiamo alcuni sottomoduli caratteristici nel caso speciale din+1 punti in posizione generale, in ℙⁿ.

     

Approximation Numbers of Operators on Normed Linear Spaces

In [1], B?ttcher et. al. showed that if T is a bounded linear operator on a separable Hilbert space H, {e_j}_j=1^￥H, \{e_{j}\}_{j=1}^{\infty} is an orthonormal basis of H and P_n is the orthogonal projection onto the span of {e_j}_j=1ⁿ\{e_{j}\}_{j=1}^{n}, then for each k ? \mathbbNk \in {\mathbb{N}}, the sequence {s_k(P_nTP_n)}\{s_{k}(P_{n}TP_{n})\} converges to s_k(T), where for a bounded operator A on H, s_k(A) denotes the kth approximation number of A, that is, s_k(A) is the distance from A to the set of all bounded linear operators of rank at most k − 1. In this paper we extend the above result to more general cases. In particular, we prove that if T is a bounded linear operator from a separable normed linear space X to a reflexive Banach space Y and if {P_n} and {Q_n} are sequences of bounded linear operators on X and Y, respectively, such that ||P_n|| ||Q_n|| ￡ 1\|P_n\| \|Q_n\| \leq 1 for all n ? \mathbbNn \in {\mathbb{N}} and {Q_nTP_n} converges to T under the weak operator topology, then {s_k(Q_nTP_n)}\{s_{k}(Q_{n}TP_{n})\} converges to s_k(T). We also obtain a similar result for the case of any normed linear space Y which is the dual of some separable normed linear space. For compact operators, we give this convergence of s_k(Q_nTP_n)s_{k}(Q_{n}TP_{n}) to s_k(T) with separability assumptions on X and the dual of Y. Counter examples are given to show that the results do not hold if additional assumptions on the space Y are removed. Under separability assumptions on X and Y, we also show that if there exist sequences of bounded linear operators {P_n} and {Q_n} on X and Y respectively such that (i) Q_nTP_nQ_{n}TP_{n} is compact, (ii) ||P_n|| ||Q_n|| ￡ 1\|P_{n}\| \|Q_{n}\| \leq 1 and (iii) {Q_nTP_n}\{Q_{n}TP_{n}\} converges to T in the weak operator topology, then {s_k(Q_nTP_n)}\{s_k(Q_{n}TP_{n})\} converges to s_k(T) if and only if s_k(T) = s_k(T￠)s_{k}(T) = s_{k}(T^\prime). This leads to a generalization of a result of Hutton [3], proved for compact operators between normed linear spaces.     

On the derivatives of X-ray functions

We show that for every n \geqq 4, 0 \leqq k \leqq n - 3, p ? (0, 3] n \geqq 4, 0 \leqq k \leqq n - 3, p \in (0, 3] and every origin-symmetric convex body K in \mathbbRⁿ \mathbb{R}^n , the function ||x ||^-k₂ ||x ||^-n+k+p_K \parallel x \parallel^{-k}_{2} \parallel x \parallel^{-n+k+p}_{K} represents a positive definite distribution on \mathbbRⁿ \mathbb{R}^n , where ||·||₂ \parallel \cdot \parallel_{2} is the Euclidean norm and ||·||_K \parallel \cdot \parallel_{K} is the Minkowski functional of K. We apply this fact to prove a result of Busemann-Petty type that the inequalities for the derivatives of order (n - 4) at zero of X-ray functions of two convex bodies imply the inequalities for the volume of average m-dimensional sections of these bodies for all 3 \leqq m \leqq n 3 \leqq m \leqq n . We also prove a sharp lower estimate for the maximal derivative of X-ray functions of the order (n - 4) at zero.     

Small zeros of additive forms in several variables

Letf(X) be an additive form defined by

Estimating a general function of a quadratic function

Let be an observation from a spherically symmetric distribution with unknown location parameter . For a general non-negative function c, we consider the problem of estimating c(||x − θ||²) under the usual quadratic loss. For p ≥ 5, we give sufficient conditions for improving on the unbiased estimator γ₀ of c(||x − θ||²) by competing estimators γ _s = γ₀ + s correcting γ₀ with a suitable function s. The main condition relies on a partial differential inequality of the form k Δs + s ² ≤ 0 for a certain constant k ≠ 0. Our approach unifies, in particular, the two problems of quadratic loss estimation and confidence statement estimation and allows to derive new results for these two specific cases. Note that we formally establish our domination results (that is, with no recourse to simulation).      

On Sikkema-Kantorovič polynomials of order K

Let a_n≥0 and F(u)∈C [0,1], Sikkema constructed polynomials: , ifα _n ≡0, then B_n (0, F, x) are Bernstein polynomials. Let , we constructe new polynomials in this paper: Q _n ^(k) (α _n ,f(t))=d ^k /dx ^k B _n+k (α _n ,F _k (u),x), which are called Sikkema-Kantorovic polynomials of order k. Ifα _n ≡0, k=1, then Q_n ⁽¹⁾ (0, f(t), x) are Kantorovič polynomials P_n(f). Ifα _n =0, k=2, then Q_n ⁽²⁾, (0, f(t), x) are Kantorovič polynomials of second order (see Nagel). The main result is: Theorem 2. Let 1≤p≤∞, in order that for every f∈L^P [0, 1], , it is sufficient and necessary that , § 1. Let f(t) de a continuous function on [a, b], i. e., f∈C [a, b], we define^{[1–2],[8–10]}: . As usual, for the space L^p [a,b](1≤p<∞), we have and L[a, b]=l¹[a, b]. Letα _n ⩾0and F(u)∈C[0,1],Sikkema-Bernstein polynomials ^{[3] [4]}. The author expresses his thanks to Professor M. W. Müller of Dortmund University at West Germany for his supports.     

Numerical verification of condition for approximately midconvex functions

Let X be a normed space and V be a convex subset of X. Let a\colon \mathbbR₊ ? \mathbbR₊{\alpha \colon \mathbb{R}_+ \to \mathbb{R}_+}. A function f \colon V ? \mathbbR{f \colon V \to \mathbb{R}} is called α-midconvex if

Approximation by the Bernstein–Durrmeyer Operator on a Simplex

For the Jacobi-type Bernstein–Durrmeyer operator M _n,κ on the simplex T ^d of ℝ ^d , we proved that for f∈L ^p (W _κ ;T ^d ) with 1<p<∞,

On approximation of periodic functions by Fourier sums

Let L _p , 1 ≤ p< ∞, be the space of 2π-periodic functions f with the norm || f ||_p = ( ò _{- p}^p | f |^p )^{1 \mathord}

An Inertial Proximal Algorithm with Dry Friction: Finite Convergence Results

Let H be a Hilbert space and A, B: H ⇉ H two maximal monotone operators. In this paper, we investigate the properties of the following proximal type algorithm: